Simultaneous decomposition of the variance by sources and subgroups – new insights

There are two types of decompositions: of linear combinations of random variables into contributions of individual variables (sources) and associations between them, and of populations into contributions of their subpopulations. Simultaneous treatment of the two types is called for, which takes into account the correlations between sources within subpopulations and between subpopulation means. The expected values of the subcomponents are derived and their sensitivity to correlations among sources within groups and among source group means is conducted. An example is provided, in which the correlations contribute 20--25% to total variability. This additional information is hidden when decompositions are not simultaneous.     

Combined survival analysis of cardiac patients by a Cox PH model and a Markov chain

The control and treatment of dyslipidemia is a major public health challenge, particularly for patients with coronary heart diseases. In this paper we propose a framework for survival analysis of patients who had a major cardiac event, focusing on assessment of the effect of changing LDL-cholesterol level and statins consumption on survival. This framework includes a Cox PH model and a Markov chain, and combines their results into reinforced conclusions regarding the factors that affect survival time. We prospectively studied 2,277 cardiac patients, and the results show high congruence between the Markov model and the PH model; both evidence that diabetes, history of stroke, peripheral vascular disease and smoking significantly increase hazard rate and reduce survival time. On the other hand, statin consumption is correlated with a lower hazard rate and longer survival time in both models. The role of such a framework in understanding the therapeutic behavior of patients and implementing effective secondary and primary prevention of heart diseases is discussed here.     

The effect of system aesthetics on trust, cooperation, satisfaction and annoyance in an imperfect automated system

Lack of system reliability has been repeatedly identified as a factor that decreases trust. However, aesthetics has an important role in the development of trust. Most of the research concerning the connection between aesthetics and trust focused on mobile commerce and websites while very little has been done in examining aesthetics in automated systems. This study integrated aesthetics manipulations into an imperfect in-vehicle automation system and focused on the power of aesthetics to decrease the negative effects of errors on trust, satisfaction, annoyance, and human-automation cooperation perceptions. Participants used the navigation system in either 100% or 85% accuracy levels with an aesthetic or non aesthetic system (4 conditions). In both aesthetic and non aesthetic systems, perceptions of trust, satisfaction and human automation cooperation were decreased in the imperfect system compared to the perfect one. However, in the annoyance rating, this trend was found only in the aesthetic system while in the non-aesthetic system no difference was found between the two levels of accuracy. This single effect may indicate upon the possibility that in automated systems aesthetics affects trust and satisfaction more moderately compared to mobile commerce applications and websites. However, more research is needed to assess this assumption.     

Teachers' Occupation‐Specific Work‐Family Conflict

To expand work‐family conflict (WFC) research to specific occupations, this study investigated how work and family generic and occupation‐specific stressors and support variables related to family interfering with work (F → W) and work interfering with family (W → F) among 230 Israeli high school teachers. Further expanding WFC research, the authors assessed WFC effects on burnout and vigor. Results indicated that W → F conflict was related to generic variables and more so to distinctive teaching characteristics (e.g., investment in student behavior and parent‐teacher relations). Both W → F and F → W predicted burnout, whereas only F → W predicted vigor. Implications for WFC research and occupational health programs are discussed.     

A methodology for quantifying the effect of missing data on decision quality in classification problems

Decision making is often supported by decision models. This study suggests that the negative impact of poor data quality (DQ) on decision making is often mediated by biased model estimation. To highlight this perspective, we develop an analytical framework that links three quality levels – data, model, and decision. The general framework is first developed at a high-level, and then extended further toward understanding the effect of incomplete datasets on Linear Discriminant Analysis (LDA) classifiers. The interplay between the three quality levels is evaluated analytically – initially for a one-dimensional case, and then for multiple dimensions. The impact is then further analyzed through several simulative experiments with artificial and real-world datasets. The experiment results support the analytical development and reveal nearly-exponential decline in the decision error as the completeness level increases. To conclude, we discuss the framework and the empirical findings, elaborate on the implications of our model on the data quality management, and the use of data for decision-models estimation.     

Factor Analysis Revisited – How Many Factors are There?

We suggest a procedure to improve the overall performances of several existing methods for determining the number of factors in factor analysis by using alternative measures of correlation: Pearson's, Spearman's, Gini's, and a robust estimator of the covariance matrix (MCD). We examine the effect of the choice of the covariance used on the number of factors chosen by the KG rule of one, the 80% rule, the Minimum average partial (MAP), and the Parallel Analysis Methodology (PAM). Extensive simulations show that when the entire (or part) of the data come from heavy-tail (lognormal) distributions, ranking the variables which come from non symmetric distributions improves the performances of the methods. In this case, Gini is slightly better than Spearman. The PAM and MAP procedures are qualitatively superior to the KG and the 80% rules in determining the true number of factors. A real example involving data on document authorship is analyzed.